sign-language-classification

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1351
Accuracy: 0.96

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0002
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.01
num_epochs: 32

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.6016	1.0	100	1.5038	0.8
1.1072	2.0	200	0.6959	0.8675
0.6195	3.0	300	0.5236	0.87
0.5559	4.0	400	0.4819	0.87
0.389	5.0	500	0.3392	0.9
0.3878	6.0	600	0.3600	0.9025
0.3309	7.0	700	0.3312	0.9075
0.3397	8.0	800	0.2596	0.9225
0.3033	9.0	900	0.2056	0.935
0.2765	10.0	1000	0.2802	0.9175
0.2846	11.0	1100	0.3276	0.9025
0.2443	12.0	1200	0.3689	0.8975
0.2682	13.0	1300	0.2805	0.915
0.2053	14.0	1400	0.2437	0.9225
0.2453	15.0	1500	0.2646	0.92
0.1896	16.0	1600	0.2489	0.925
0.1841	17.0	1700	0.2393	0.9275
0.1406	18.0	1800	0.1935	0.945
0.1573	19.0	1900	0.2544	0.92
0.155	20.0	2000	0.1940	0.9475
0.1563	21.0	2100	0.2021	0.9325
0.133	22.0	2200	0.2413	0.9325
0.117	23.0	2300	0.1939	0.9375
0.1455	24.0	2400	0.1685	0.9575
0.144	25.0	2500	0.1787	0.9475
0.1119	26.0	2600	0.1511	0.96
0.1053	27.0	2700	0.1308	0.965
0.0964	28.0	2800	0.1042	0.9725
0.0938	29.0	2900	0.1751	0.9425
0.0881	30.0	3000	0.1066	0.965
0.0854	31.0	3100	0.1116	0.97
0.1002	32.0	3200	0.1351	0.96

Framework versions

Transformers 4.38.2
Pytorch 2.1.2
Datasets 2.18.0
Tokenizers 0.15.2